From the Edge to the Core: Scaling Networks to Support Exascale Computing and Data
Inder Monga, Eli Dart, ESnet/Lawrence Berkeley National Laboratory
Tue 4:00 – 4:45pm
Big-data science is an end-to-end problem – from data generation at science instruments i.e. experimental and observational facilities to on-demand data movement to the distributed set of available computational elements. The role of networking is key to this workflow. In addition, as data volumes grow exponentially, high-performance architecture of network and end-systems is critical for science. In this talk, ESnet discusses key challenges to exascale data and compute, end-to-end performance and architecture of DOE computational facilities, advances in optical networking and architectures being considered for ESnet upgrade to its next-generation network.